Color accuracy of SQ patterns

For all topics which do not fit in the other categories.

Moderators: rcperryls, Rose, karen4bells, Serinde, Alex

Post Reply
User avatar
SteveM
Posts: 242
Joined: Mon Jul 20, 2015 6:31 am
Location: Oregon

Color accuracy of SQ patterns

Post by SteveM »

I recently purchased the Scarlet Quince pattern of Washington crossing the Delaware River and am beginning to doubt the accuracy of the colors in the pattern to the original art. While reviewing the floss list I noticed an abundance of brown tones and very little red which I found surprising considering that most of the prominent figures in the picture had some red on them. To get to the bottom of this I purchased a few skeins of floss and stitched up the red portion of Washinton's cloak 2 over 1 on a scrap of 25 count white Lugana. I also used the Pattern Maker feature of blending threads to replicate what it thought the cloak should look like based on the SQ pattern. Here are my results.

Original artwork found on Wikimedia and scaled to the same size as the pattern:
Image

A close up of the SQ pattern's image representing what the finished stitch will look like:
Image

Ok, a little shifted to brown but not too bad. That's a pretty good representation of the original art colors so let's see what Pattern Maker predicts using the SQ pattern:
Image

Oh gross, let's hope PM has some floss errors or doesn't know how to blend DMC threads very well. The proof is in the pudding and here is my stitched result:
Image

Thankfully the real result isn't too bad compared to the PM prediction, but it still isn't as red as the SQ image predicts nor as red as the original art. I suppose if I didn't know any better I'd be happy with the colors and the excellent detail and blending. But my concern is that I'd ultimately end up with a brown sky and brown water and a brown cast to everything.

I've seen Allyn's completed Vanitas Still Life and compared it to the SQ image and feel that they are a pretty close match. However, that was a pretty earth toned image to begin with so a brown cast to the colors is appropriate. What do you all think? Am I expecting too much of the limited colors available and/or should I just trust SQ and proceed? Will it look better with the other colors around it to provide some contrast? All comments and suggestions are appreciated.
-Steve

__________________________________________________________________
WIPs: HAED "Mushroom Inn" & "The Ionian Mission"
Finishes: Dim. Gold "Woodland Winter" & HAED "SK History of Chocolate"
User avatar
Allyn
Posts: 2226
Joined: Sat Mar 30, 2013 4:57 pm
Location: Mississippi Gulf Coast
Contact:

Re: Color accuracy of SQ patterns

Post by Allyn »

Most folks wouldn't be that particular and scrutinize the colors that closely. I've seen some really bad color conversions -- mostly from HaED -- and folks are perfectly happy with the results. I'm a scrutinizer, though. I find that if the chart doesn't use blends, I'm usually not satisfied when looking at converted artwork. I see the differences and I'm not happy with them. Even in Vanitas Still Life, I was not completely happy. I know I could do a better job converting the image to my satisfaction. Here's what I find: commercially produced charts limit the number of colors to something most stitchers would find manageable. In Vanitas, that was 75 colors. When I tinker with an image, the least number of colors that I find make a satisfactory conversion is 120, usually around 160 and with the blends, I need at least 150 to 200 different DMC colors. Most folks would find that too intimidating or so unmanageable, they wouldn't do it. So to get a chart palette down to a manageable size, the chart producers limit the number of colors. The overall image looks fine, but when you start comparing individual colors, the charted image will always look less than ideal. Also consider that you are comparing the charted version to an image from a different source. I can pull that same image from five different sources and the colors -- the reds, for example -- could be slightly different in each image.

That being said, there were many times while I was stitching Vanitas that I thought, "Oh crap, this is wrong. This doesn't look right. Oh <bleep> this is ruined." But when I finished the section and stood back to look at it, everything was fine. The colors worked. Five inches from my nose it looked like a hodge-podge of weird shading, but when I stood back and viewed it, the blends blended and the image was fine. Many of those "browns" aren't brown at all; they're blends of reds and greens, blues and golds, and purples and golds.

So. . .continue with it. Scrutinizing small sections and fretting over individual colors will make you crazy. When you stand back and look at the whole image, it'll look fine.
_________________________________________________________

WIP
Image
User avatar
SteveM
Posts: 242
Joined: Mon Jul 20, 2015 6:31 am
Location: Oregon

Re: Color accuracy of SQ patterns

Post by SteveM »

Allyn,

You are quite right that there are other images out there with slightly different color balances to them and I've no idea which is more accurate.

I've been plying around with the image in Pattern Maker and Richard's software and have a hard time getting the same level of detail and blending of colors as the SQ pattern so I do appreciate that she put many hours into the pattern. I am curious which software you have been using and what is the source of your blended colors.
-Steve

__________________________________________________________________
WIPs: HAED "Mushroom Inn" & "The Ionian Mission"
Finishes: Dim. Gold "Woodland Winter" & HAED "SK History of Chocolate"
User avatar
Allyn
Posts: 2226
Joined: Sat Mar 30, 2013 4:57 pm
Location: Mississippi Gulf Coast
Contact:

Re: Color accuracy of SQ patterns

Post by Allyn »

When I did Vanitas, I used the colors in the SQ chart. There were a couple that I would have liked to "reblend.'' I wish I'd had Richard's color picker when I started it or I would have; manually doing the calculations to find a better blend was too much trouble. I went with SQ's blends. The ones I would have reblended were a couple that had a very light shade paired with a very dark shade -- like, for example, if you paired ecru with 3371. In heavy confetti areas, it wasn't a problem, but in areas of the background where there were small fields of that symbol, the colors look a bit stripey instead of being a smooth blend.

Imagine for example, that white and black paired together gives the perfect "blend" as far as the calculations go, but you know the extreme light color and extreme dark color will never make a smooth blend. If you take 413 and 415, it makes a color that isn't as close in calculation, but it's close enough and you know the two grays are close enough to make a smooth blend. That's what I would have changed for a couple of SQs blends.

When I tinker with an image now, it's usually Richard's program because of the ease with which it incorporates blends. Converting art to stitch without blends doesn't get the colors close enough for my satisfaction. I also use PCStitch10 and Cross Stitch Professional Platinum (DP Software). Of the two, I prefer CCPP over PCStitch, but dealing with blends still isn't as easy as with Richard's program. I had at one time tried Pattern Maker, but it didn't hold my interest when compared to other programs I had available. I don't remember if it was disatisfaction with color conversion or if it lacked other tools that I wanted.

So, I use Richard's program or I use Richard's color picker with the other XS programs I have. There are some tricks to prepare an image for conversion. Just importing it into the software with no prep doesn't seem to get the best results when you compare the before and after.
_________________________________________________________

WIP
Image
User avatar
Allyn
Posts: 2226
Joined: Sat Mar 30, 2013 4:57 pm
Location: Mississippi Gulf Coast
Contact:

Re: Color accuracy of SQ patterns

Post by Allyn »

SteveM wrote: ... While reviewing the floss list I noticed an abundance of brown tones and very little red which I found surprising considering that most of the prominent figures in the picture had some red on them. ...
And just a side note, when I looked at the floss list for Vanitas, I saw some browns and tans, but there were a lot of what I'd call Christmas colors and Easter colors. In looking at the model, I thought, "what in the heck do I use those for?" Those reds, greens, blues and purples -- some in pastels and some in vivid colors -- became browns and tans in blends.

Unlike some of the art-to-stitch conversions that lose details because areas of dark shading end up being large areas of solid black after the conversion, SQ uses alot of brown shades and browns in blends to bring out the details in the dark areas. No matter how dark an area was in Vanitas, there was never even a single stitch that was just black. I think you'll find that's where most of the browns are going to be used -- in dark details.

Is there a symbol in the chart that is just black (310) with no other color blended with it? I'd be willing to bet there is not.
_________________________________________________________

WIP
Image
User avatar
richardandtracy
Posts: 5419
Joined: Wed May 28, 2014 4:27 pm
Location: Kent, UK
Contact:

Re: Color accuracy of SQ patterns

Post by richardandtracy »

Have to confess I am looking at this thread with interest, but can't contribute too much - I have less than a year's experience, and only one completed piece, in higher colour count stitching. Not even used blending yet - I'll be ordering the threads for my first attempt tomorrow.

The only thing I can say is individual colours can look very different from the finished section. In the chart of my honeymoon picture, much the side of my wife's nostril was made up from two dithered shades of red (bright crimson and scarlet - can't remember the numbers at the moment, my chart is elsewhere). Anyway, when the surrounding picture was completed, the area looked brown, even though the threads making it up certainly were not. Furthermore, in the teeth there were colours including tangerine orange and lime green - completely mad colours if you were trying to think of them yourself, but the teeth look really teeth-like.

Not sure I can say anything further apart from 'It could well come out right in the end', but I suspect there is no guarantee that it'll be as good as you want. I wish there was an easy solution [I mean, apart from the obvious of lowering your expectations!], but I think it will probably involve you almost re-charting the areas you are not happy with. This then leads into a potential serious risk of the re-charted area not having a hue that matches the rest of the image. Allyn has a good deal of experience on this - but I suspect using the thread picker program to get a revised colour for individual stitches was hard work. A reason for Allyn's dissatisfaction with some of the SQ blends in her Vanita's is, I think, with some commercial software is that the software assumes that because it can be done mathematically, the blend will always work. I think there must be a brightness test, saturation test and hue test too to determine if the blend will work. In my software, I have tried to keep the hue difference to below 45 degrees on the colour wheel (max 180), and saturation/brightness to below 0.26 (max 1.0). This way DMC 3371 and 310 produce a blend, but many other possible combinations are excluded out of the potential 100,000. Nonetheless, this allows around 12000 blends, which should be enough for most 'discerning' users. If it isn't... Oh help!

If you can think of a process that you regularly go through to make the chart better, then maybe it could be automated or made semi-automatic, so it would be quicker. Let me know & I'll see what I can do to help. For example: if you know at one stitch the colour has to be black, and (say) two stitches up and five to the right it has to be scarlet, I could write a little colour picker to choose the intervening blends by location. That sort of thing is possible, and could save a lot of time if that is the way you work. (The engineer in me wants to get tools to help the way I work, not force a change in working pattern unless it's obvious it's much easier.. Which is one reason why I find Microsoft Word so difficult to use, for all its interface flexibility, it makes on-the-fly formatting using the keyboard on shared documents much more difficult than it needs to be. Ooops. Must get off my soapbox.)

Finally, I suspect those who are very, very picky about how their picture will look will almost always be forced to chart their own image so that it is exactly right. That way the compromises inherent in having a chart suitable for a wider audience than 1 will be avoided.

Regards,

Richard
http://www.chestnutpens.co.uk
Free Charting Program for PC's Info Zip Installation
User avatar
SteveM
Posts: 242
Joined: Mon Jul 20, 2015 6:31 am
Location: Oregon

Re: Color accuracy of SQ patterns

Post by SteveM »

Allyn wrote:Is there a symbol in the chart that is just black (310) with no other color blended with it? I'd be willing to bet there is not.
Of the 121 colors used only 14 are solids and none of those is black or white. Ecru is the lightest solid color. I had already checked for solid black because PM tends to take dark areas and make them solid black even when I feel that there are suitable dark browns available. If you crank up brightness for the whole image to correct the dark areas the light areas go white. So I really like that the pattern has no wholly black or white stitches.

Richard,
I'm no expert stitcher myself and have no idea how to hand edit a pattern for best result. My main reason for exploring charting my own pattern is a lack of patterns that I find interesting to me and a desire to have a larger amount of detail than often available. The best tools that I have now are PM and your program. I find that PM is very fast and tends to deliver a decent result with few colors, but adding blended threads is a very slow process and I don't think there is a way to create a 12,000 color palette of blends to use while importing a photo. Your program gives me the most control and flexibility, but for some reason seems to need more colors to get a good looking result but also offers the best looking result especially with the new custom dither feature. I'll post an example of Washington's cloak using your program and the number of colors used in the pattern as a comparison.
-Steve

__________________________________________________________________
WIPs: HAED "Mushroom Inn" & "The Ionian Mission"
Finishes: Dim. Gold "Woodland Winter" & HAED "SK History of Chocolate"
User avatar
richardandtracy
Posts: 5419
Joined: Wed May 28, 2014 4:27 pm
Location: Kent, UK
Contact:

Re: Color accuracy of SQ patterns

Post by richardandtracy »

That will be interesting.
Not sure why my program ends up with more colours, but it seems to. Maybe need to aim for a slightly higher number and then merge some of the ones with lower stitch numbers. I need to re-assess how it offers alternatives - it should be a bit more intelligent about the selection in the 'merge colours' window.

Regards,

Richard.
http://www.chestnutpens.co.uk
Free Charting Program for PC's Info Zip Installation
User avatar
SteveM
Posts: 242
Joined: Mon Jul 20, 2015 6:31 am
Location: Oregon

Re: Color accuracy of SQ patterns

Post by SteveM »

I hesitate to revive this thread, but something very interesting has just happened. Yesterday Scarlet Quince released an online tool for viewing and marking up your purchased SQ patterns. For the first time that I am aware of this now provides a color view of the chart's key. This gives me the information I need to determine if I think the SQ color definitions match my floss in hand.

My complaint was that George Washington's cape was rather brown when I stitched it up rather than the muted red expected. I even replicated that portion of the pattern in Pattern Maker to confirm that the floss colors that I used would result in a brown cape. Now that I have compared my floss colors to the color key I can see that the SQ key is a pretty good match for most colors, but there is a tendency for the SQ color to be slightly redder than expected.

However, DMC 300 and 919 were significantly off. My two samples of DMC 300 are pretty much a medium brown, but the SQ key shows a dark terracotta color with plenty of red in it. And the DMC 919 that I have is what I would call a medium shade of terracotta but the SQ key borders on dark cherry red. I tweaked my Pattern Maker colors for just DMC 300 to match the SQ color and the resulting changes in the cape made a remarkable shift to the red that it should be.

In summary, if I were to stitch an SQ pattern I'd first use their color key of the chart to re-map colors as needed to better matches. For example, where SQ calls for DMC 300 I'd use 3777. Where DMC 919 is called for I'd use 304.
-Steve

__________________________________________________________________
WIPs: HAED "Mushroom Inn" & "The Ionian Mission"
Finishes: Dim. Gold "Woodland Winter" & HAED "SK History of Chocolate"
User avatar
Allyn
Posts: 2226
Joined: Sat Mar 30, 2013 4:57 pm
Location: Mississippi Gulf Coast
Contact:

Re: Color accuracy of SQ patterns

Post by Allyn »

As far as I can tell, all SQ patterns are made from public domain images. Instead of doing that work to correct colors in their chart, wouldn't you rather chart the image yourself in Richard's program?
_________________________________________________________

WIP
Image
User avatar
SteveM
Posts: 242
Joined: Mon Jul 20, 2015 6:31 am
Location: Oregon

Re: Color accuracy of SQ patterns

Post by SteveM »

Knowing what I know now and with the improvements that Richard has made to his program - I think I would chart it myself. Maybe I can do that in 10 years when I've finished my current WIPs.
-Steve

__________________________________________________________________
WIPs: HAED "Mushroom Inn" & "The Ionian Mission"
Finishes: Dim. Gold "Woodland Winter" & HAED "SK History of Chocolate"
User avatar
richardandtracy
Posts: 5419
Joined: Wed May 28, 2014 4:27 pm
Location: Kent, UK
Contact:

Re: Color accuracy of SQ patterns

Post by richardandtracy »

Must warn you that my program is still using Public Domain colour definitions for DMC that seem traceable back to the now defunct xstitchtreasures.com and some references suggest they were an unofficial DMC release prior to that. They seem to match PM and PCStitch definitions except where I discovered errors the hard way myself & corrected them.

My plan is to create my own source data, but have not yet had an opportunity to photograph my floss card outside on a dull day. Only in the last week has the weather got to the stage where I could, and with such a late spring there are a lot of more pressing jobs to be done outside.

It is on the 'To Do' list though.

Regards,

Richard.
http://www.chestnutpens.co.uk
Free Charting Program for PC's Info Zip Installation
User avatar
SteveM
Posts: 242
Joined: Mon Jul 20, 2015 6:31 am
Location: Oregon

Re: Color accuracy of SQ patterns

Post by SteveM »

I don't think they have to be perfect, especially when you consider possible changes between lots and over time as the chemicals and process change. However, DMC 300 seems to be poorly represented by some.

SQ is using 118,34,0 which is a bit too red.
HAED and Artecy are using the PM default of 120,78,36 which is a bit too medium brown.
Your definition is 111,47,0 which you might think is going to look like SQ, but the changes are just enough to tone the red down to a reasonable approximation of the floss.
Pic2Pat is also using 111,47,0.
MacStitch tends to do pretty good conversions and they are using 119,62,43 which is a tad bit better than HAED/PM, but over all your is the closest of these all.

Just playing around and guessing, I think that 107,45,32 would be fairly close for the skein that I have (when viewed under bright indoor light). I have a PDF color card from DMC and even there it looks a bit too dark and red, but the definition you are using would still be the best match for it.

Around here that clearing in the weather that you are waiting for is called "June".
-Steve

__________________________________________________________________
WIPs: HAED "Mushroom Inn" & "The Ionian Mission"
Finishes: Dim. Gold "Woodland Winter" & HAED "SK History of Chocolate"
Post Reply